Class: TokenTrieNER

Inherits:
NER
  • Object
show all
Defined in:
lib/rbbt/ner/token_trieNER.rb

Defined Under Namespace

Modules: EnumeratedArray Classes: Code

Instance Attribute Summary collapse

Class Method Summary collapse

Instance Method Summary collapse

Methods inherited from NER

#entities, #extract

Constructor Details

#initialize(type = nil, file = nil, options = {}) ⇒ TokenTrieNER

Returns a new instance of TokenTrieNER.



257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 257

def initialize(type = nil, file = nil, options = {})
  options = Misc.add_defaults options, :longest_match => true, :no_clean => false, :slack => nil, :split_at => nil,
    :persist => false
  @slack = slack
  @longest_match = options.delete :longest_match
  @split_at = options.delete :split_at
  @no_clean = options.delete :no_clean
  @stem = options.delete :stem

  file = [] if file.nil?
  file = [file] unless Array === file
  persist_options = Misc.pull_keys options, :persist
  @index = Persist.persist_tsv(file, options, persist_options) do |data|
    data.serializer = :marshal if data.respond_to? :serializer and data.serializer == :type

    @index = data
    file.each do |f| 
      merge(f, type)
    end

    @index
  end
end

Instance Attribute Details

#index ⇒ Object

Returns the value of attribute index.



256
257
258
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 256

def index
  @index
end

#longest_match ⇒ Object

Returns the value of attribute longest_match.



256
257
258
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 256

def longest_match
  @longest_match
end

#no_clean ⇒ Object

Returns the value of attribute no_clean.



256
257
258
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 256

def no_clean
  @no_clean
end

#slack ⇒ Object

Returns the value of attribute slack.



256
257
258
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 256

def slack
  @slack
end

#split_at ⇒ Object

Returns the value of attribute split_at.



256
257
258
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 256

def split_at
  @split_at
end

#stem ⇒ Object

Returns the value of attribute stem.



256
257
258
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 256

def stem
  @stem
end

#type ⇒ Object

Returns the value of attribute type.



256
257
258
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 256

def type
  @type
end

Class Method Details

.clean(token, stem = false) ⇒ Object 



8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 8

def self.clean(token, stem = false)
  if token.length > 3
    upcase = token !~ /[a-z]/
    token = token.downcase.sub(/-/,'')

    if stem && ! upcase
      require 'stemmer'
      if stem == :double
        token = token.stem.stem
      else
        token = token.stem
      end
    end

    token
  else
    token
  end
end

.find(index, tokens, longest_match = true, slack = nil, first = true) ⇒ Object 



213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 213

def self.find(index, tokens, longest_match = true, slack = nil, first = true)
  head = tokens.next
  
  next_index = follow(index, head)


  return find_fail(index, tokens, head, longest_match, slack, first) if next_index.nil?

  if not tokens.left?
    if next_index.include? :END
      return [next_index[:END], [head]]
    else
      return find_fail(index, tokens, head, longest_match, slack, first)
    end
  else

    return [next_index[:END], [head]] if next_index.include?(:END) and not longest_match

    matches = find(next_index, tokens, longest_match, slack, false) # Recursion

    if not matches.nil?
      matches.last.unshift head
      return matches
    end
    
    return [next_index[:END], [head]] if next_index.include?(:END)

    return find_fail(index, tokens, head, longest_match, slack, first)
  end
end

.find_fail(index, tokens, head, longest_match, slack, first) ⇒ Object 



200
201
202
203
204
205
206
207
208
209
210
211
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 200

def self.find_fail(index, tokens, head, longest_match, slack, first)
  if Proc === slack and not first and not head.nil? and tokens.left? and slack.call(head) 
    matches = find(index, tokens, longest_match, slack, false) # Recursion
    if not matches.nil?
      matches.last.unshift head
      return matches
    end
  end

  tokens.back
  return nil
end

.follow(index, head) ⇒ Object

{{{ Matching



184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 184

def self.follow(index, head)
  res = nil

  if index.include? head
    return index[head]
  end

  return nil unless (not TokyoCabinet::HDB === index ) and index.include? :PROCS

  index[:PROCS].each do |key,value|
    return value if key.call(head)
  end

  nil
end

.index_for_tokens(tokens, code, type = nil, slack = nil) ⇒ Object 



109
110
111
112
113
114
115
116
117
118
119
120
121
122
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 109

def self.index_for_tokens(tokens, code, type = nil, slack = nil)
  if not tokens.left?
    {:END => [Code.new(code, type)]}
  else
    head = tokens.next
    if (slack.nil? or not slack.call(head))
      res = {head => index_for_tokens(tokens, code, type, slack)}
    else
      res = {head => index_for_tokens(tokens, code, type, slack)}.merge(index_for_tokens(tokens, code, type, slack))
    end
    tokens.back
    res
  end
end

.make_match(match_tokens, type, codes) ⇒ Object 



244
245
246
247
248
249
250
251
252
253
254
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 244

def self.make_match(match_tokens, type, codes)
  match = ""
  match_offset = match_tokens.first.offset
  match_tokens.each{|t| 
    match << " " * (t.offset - (match_offset + match.length)) if t.offset > (match_offset + match.length)
    match << ((t.respond_to?(:original) and not t.original.nil?) ? t.original : t)
  }

  type = type.first
  NamedEntity.setup(match, :offset => match_tokens.first.offset, :entity_type => type, :code => codes, :type => type)
end

.merge(index1, index2) ⇒ Object 



124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 124

def self.merge(index1, index2)
  index1.write if index1.respond_to? :write and not index1.write?
  index2.each do |key, new_index2|
    case
    when key == :END
      end1 = index1[:END] || []
      end1 += new_index2.reject{|new| end1.collect{|e| e.to_s }.include? new.to_s }
      end1.uniq!
      index1[:END] = end1
    when index1.include?(key)
      new = merge(index1[key], new_index2)
      index1[key] = new
    else
      index1[key] = new_index2
    end
  end
  index1.read if index1.respond_to? :read

  index1
end

.prepare_token(token, start, extend_to_token = true, no_clean = false, stem = false) ⇒ Object 



28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 28

def self.prepare_token(token, start, extend_to_token = true, no_clean = false, stem = false)
  if no_clean
    if extend_to_token
      Token.setup(token, :offset => start, :original => token)
    else
      token
    end
  else
    if extend_to_token
      Token.setup(clean(token, stem), :offset => start, :original => token)
    else
      clean(token, stem)
    end
  end
end

.process(index, hash, type = nil, slack = nil, split_at = nil, no_clean = false, stem = false) ⇒ Object 



145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 145

def self.process(index, hash, type = nil, slack = nil, split_at = nil, no_clean = false, stem = false)

  chunk_size = hash.size / 100
  items_in_chunk = 0
  tmp_index = {}
  hash.send(hash.respond_to?(:through)? :through : :each) do |code, names|
    names = Array === names ? names : [names]
    names.flatten! if Array === names.first and not Segment === names.first.first

    if names.empty?
      names.unshift code unless TSV === hash and not (hash.fields.nil? or hash.fields.empty?)
    end

    names.each do |name|
      next if name.empty? or (String === name and name.length < 2)

      tokens = Array === name ? name : tokenize(name, false, split_at, no_clean, stem) 
      tokens.extend EnumeratedArray

      token_index = index_for_tokens(tokens, code, type, slack)

      tmp_index = merge(tmp_index, token_index) unless tokens.empty?

      items_in_chunk += 1

      if items_in_chunk > chunk_size
        index = merge(index, tmp_index)
        tmp_index = {}
        items_in_chunk = 0
      end
    end
  end
  index = merge(index, tmp_index)

  index
end

.tokenize(text, extend_to_token = true, split_at = nil, no_clean = false, stem = false, start = 0) ⇒ Object 



44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 44

def self.tokenize(text, extend_to_token = true, split_at = nil, no_clean = false, stem = false, start = 0)
  split_at = /\s|(\(|\)|[-."':,;])/ if split_at.nil?

  tokens = []
  while matchdata = text.match(split_at)
    tokens << prepare_token(matchdata.pre_match, start, extend_to_token, no_clean, stem) unless matchdata.pre_match.empty?
    tokens << prepare_token(matchdata.captures.first, start + matchdata.begin(1), extend_to_token, no_clean, stem) if matchdata.captures.any? and not matchdata.captures.first.empty?
    start += matchdata.end(0)
    text = matchdata.post_match
  end
   
  tokens << prepare_token(text, start, extend_to_token, no_clean, stem) unless text.empty?

  tokens
end

Instance Method Details

#match(text) ⇒ Object 



307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 307

def match(text)
  tokens = Array === text ? text : TokenTrieNER.tokenize(text, true, split_at, no_clean, stem)

  tokens.extend EnumeratedArray
  tokens.pos = 0
  
  matches = []
  while tokens.left?
    new_matches = TokenTrieNER.find(@index, tokens, longest_match, slack) 

    if new_matches
      codes, match_tokens = new_matches
      matches << TokenTrieNER.make_match(match_tokens, codes.collect{|c| c.type}, codes.collect{|c| c.code})
    else
      tokens.advance
    end
  end

  matches
end

#merge(new, type = nil) ⇒ Object 



281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
 
# File 'lib/rbbt/ner/token_trieNER.rb', line 281

def merge(new, type = nil)
  case
  when TokenTrieNER === new
    Log.debug "TokenTrieNER merging other TokenTrieNER"
    TokenTrieNER.merge(@index, new.index)
  when TSV === new
    Log.debug "TokenTrieNER merging TSV"
    new.with_unnamed do
      new.with_monitor({:step => 1000, :desc => "Processing TSV into TokenTrieNER"}) do
        TokenTrieNER.process(@index, new, type, slack, split_at, no_clean, stem)
      end
    end
  when Hash === new
    Log.debug "TokenTrieNER merging Hash"
    TokenTrieNER.merge(@index, new)
  when String === new
    Log.debug "TokenTrieNER merging file: #{ new }"
    new = TSV.open(new, :flat)
    new.with_unnamed do
      new.with_monitor({:step => 1000, :desc => "Processing TSV into TokenTrieNER"}) do
        TokenTrieNER.process(@index, new, type, slack, split_at, no_clean, stem)
      end
    end
  end
end